Projet Final Python for Data Analysis

Anne-Solène IZARD Christelle LE VAN KAHN Lily LA MORELLA

I- Nettoyage du Dataset

IMPORTS

import pandas as pd
import numpy as np
import plotly.express as px
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.graph_objects as go
import plotly.figure_factory as ff
import warnings

C:\Users\lilyj\anaconda3\Lib\site-packages\paramiko\transport.py:219: CryptographyDeprecationWarning: Blowfish has been deprecated
  "class": algorithms.Blowfish,

Dataset

diabetic = pd.read_csv("diabetic_data.csv", sep=',')
#Afficher les premières lignes du Dataset
diabetic.head()

	encounter_id	patient_nbr	race	gender	age	weight	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	...	citoglipton	insulin	glyburide-metformin	glipizide-metformin	glimepiride-pioglitazone	metformin-rosiglitazone	metformin-pioglitazone	change	diabetesMed	readmitted
0	2278392	8222157	Caucasian	Female	[0-10)	?	6	25	1	1	...	No	No	No	No	No	No	No	No	No	NO
1	149190	55629189	Caucasian	Female	[10-20)	?	1	1	7	3	...	No	Up	No	No	No	No	No	Ch	Yes	>30
2	64410	86047875	AfricanAmerican	Female	[20-30)	?	1	1	7	2	...	No	No	No	No	No	No	No	No	Yes	NO
3	500364	82442376	Caucasian	Male	[30-40)	?	1	1	7	2	...	No	Up	No	No	No	No	No	Ch	Yes	NO
4	16680	42519267	Caucasian	Male	[40-50)	?	1	1	7	1	...	No	Steady	No	No	No	No	No	Ch	Yes	NO

5 rows × 50 columns

#Taille du Dataset
diabetic.shape

(101766, 50)

Avant de nettoyer, nous remplaçons les "?" par des NaN et vérifions les 0

diabetic = diabetic.replace("?",np.nan)
len(diabetic.select_dtypes('O').columns)

37

#Description des colonnes numériques
diabetic.describe()

	encounter_id	patient_nbr	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	num_lab_procedures	num_procedures	num_medications	number_outpatient	number_emergency	number_inpatient	number_diagnoses
count	1.017660e+05	1.017660e+05	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000	101766.000000
mean	1.652016e+08	5.433040e+07	2.024006	3.715642	5.754437	4.395987	43.095641	1.339730	16.021844	0.369357	0.197836	0.635566	7.422607
std	1.026403e+08	3.869636e+07	1.445403	5.280166	4.064081	2.985108	19.674362	1.705807	8.127566	1.267265	0.930472	1.262863	1.933600
min	1.252200e+04	1.350000e+02	1.000000	1.000000	1.000000	1.000000	1.000000	0.000000	1.000000	0.000000	0.000000	0.000000	1.000000
25%	8.496119e+07	2.341322e+07	1.000000	1.000000	1.000000	2.000000	31.000000	0.000000	10.000000	0.000000	0.000000	0.000000	6.000000
50%	1.523890e+08	4.550514e+07	1.000000	1.000000	7.000000	4.000000	44.000000	1.000000	15.000000	0.000000	0.000000	0.000000	8.000000
75%	2.302709e+08	8.754595e+07	3.000000	4.000000	7.000000	6.000000	57.000000	2.000000	20.000000	0.000000	0.000000	1.000000	9.000000
max	4.438672e+08	1.895026e+08	8.000000	28.000000	25.000000	14.000000	132.000000	6.000000	81.000000	42.000000	76.000000	21.000000	16.000000

#Informations générales
diabetic.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 101766 entries, 0 to 101765
Data columns (total 50 columns):
 #   Column                    Non-Null Count   Dtype 
---  ------                    --------------   ----- 
 0   encounter_id              101766 non-null  int64 
 1   patient_nbr               101766 non-null  int64 
 2   race                      99493 non-null   object
 3   gender                    101766 non-null  object
 4   age                       101766 non-null  object
 5   weight                    3197 non-null    object
 6   admission_type_id         101766 non-null  int64 
 7   discharge_disposition_id  101766 non-null  int64 
 8   admission_source_id       101766 non-null  int64 
 9   time_in_hospital          101766 non-null  int64 
 10  payer_code                61510 non-null   object
 11  medical_specialty         51817 non-null   object
 12  num_lab_procedures        101766 non-null  int64 
 13  num_procedures            101766 non-null  int64 
 14  num_medications           101766 non-null  int64 
 15  number_outpatient         101766 non-null  int64 
 16  number_emergency          101766 non-null  int64 
 17  number_inpatient          101766 non-null  int64 
 18  diag_1                    101745 non-null  object
 19  diag_2                    101408 non-null  object
 20  diag_3                    100343 non-null  object
 21  number_diagnoses          101766 non-null  int64 
 22  max_glu_serum             101766 non-null  object
 23  A1Cresult                 101766 non-null  object
 24  metformin                 101766 non-null  object
 25  repaglinide               101766 non-null  object
 26  nateglinide               101766 non-null  object
 27  chlorpropamide            101766 non-null  object
 28  glimepiride               101766 non-null  object
 29  acetohexamide             101766 non-null  object
 30  glipizide                 101766 non-null  object
 31  glyburide                 101766 non-null  object
 32  tolbutamide               101766 non-null  object
 33  pioglitazone              101766 non-null  object
 34  rosiglitazone             101766 non-null  object
 35  acarbose                  101766 non-null  object
 36  miglitol                  101766 non-null  object
 37  troglitazone              101766 non-null  object
 38  tolazamide                101766 non-null  object
 39  examide                   101766 non-null  object
 40  citoglipton               101766 non-null  object
 41  insulin                   101766 non-null  object
 42  glyburide-metformin       101766 non-null  object
 43  glipizide-metformin       101766 non-null  object
 44  glimepiride-pioglitazone  101766 non-null  object
 45  metformin-rosiglitazone   101766 non-null  object
 46  metformin-pioglitazone    101766 non-null  object
 47  change                    101766 non-null  object
 48  diabetesMed               101766 non-null  object
 49  readmitted                101766 non-null  object
dtypes: int64(13), object(37)
memory usage: 38.8+ MB

#Valeurs manquantes
diabetic.isnull().sum()

encounter_id                    0
patient_nbr                     0
race                         2273
gender                          0
age                             0
weight                      98569
admission_type_id               0
discharge_disposition_id        0
admission_source_id             0
time_in_hospital                0
payer_code                  40256
medical_specialty           49949
num_lab_procedures              0
num_procedures                  0
num_medications                 0
number_outpatient               0
number_emergency                0
number_inpatient                0
diag_1                         21
diag_2                        358
diag_3                       1423
number_diagnoses                0
max_glu_serum                   0
A1Cresult                       0
metformin                       0
repaglinide                     0
nateglinide                     0
chlorpropamide                  0
glimepiride                     0
acetohexamide                   0
glipizide                       0
glyburide                       0
tolbutamide                     0
pioglitazone                    0
rosiglitazone                   0
acarbose                        0
miglitol                        0
troglitazone                    0
tolazamide                      0
examide                         0
citoglipton                     0
insulin                         0
glyburide-metformin             0
glipizide-metformin             0
glimepiride-pioglitazone        0
metformin-rosiglitazone         0
metformin-pioglitazone          0
change                          0
diabetesMed                     0
readmitted                      0
dtype: int64

#pourcentage de valeurs nulles
diabetic.isnull().sum()[diabetic.isnull().sum()>1]/len(diabetic)*100

race                  2.233555
weight               96.858479
payer_code           39.557416
medical_specialty    49.082208
diag_1                0.020636
diag_2                0.351787
diag_3                1.398306
dtype: float64

We decide to delete the columns with a Null rate exceeding 35%.

The 'weight' column has over 96% missing values.

The 'payer_code' column has over 39%.

And the 'medical_speciality' column has over 49%.

#Valeurs dupliquées
diabetic.duplicated().sum()

0

#Valeurs uniques par colonnes
diabetic.nunique()

encounter_id                101766
patient_nbr                  71518
race                             5
gender                           3
age                             10
weight                           9
admission_type_id                8
discharge_disposition_id        26
admission_source_id             17
time_in_hospital                14
payer_code                      17
medical_specialty               72
num_lab_procedures             118
num_procedures                   7
num_medications                 75
number_outpatient               39
number_emergency                33
number_inpatient                21
diag_1                         716
diag_2                         748
diag_3                         789
number_diagnoses                16
max_glu_serum                    4
A1Cresult                        4
metformin                        4
repaglinide                      4
nateglinide                      4
chlorpropamide                   4
glimepiride                      4
acetohexamide                    2
glipizide                        4
glyburide                        4
tolbutamide                      2
pioglitazone                     4
rosiglitazone                    4
acarbose                         4
miglitol                         4
troglitazone                     2
tolazamide                       3
examide                          1
citoglipton                      1
insulin                          4
glyburide-metformin              4
glipizide-metformin              2
glimepiride-pioglitazone         2
metformin-rosiglitazone          2
metformin-pioglitazone           2
change                           2
diabetesMed                      2
readmitted                       3
dtype: int64

Examide and citoglipton contain only one value in their column, we will drop them. Let's check the columns with 2 different values.

for column in diabetic.columns:
    if diabetic[column].nunique() == 2:
        print(diabetic[column].value_counts())

No        101765
Steady         1
Name: acetohexamide, dtype: int64
No        101743
Steady        23
Name: tolbutamide, dtype: int64
No        101763
Steady         3
Name: troglitazone, dtype: int64
No        101753
Steady        13
Name: glipizide-metformin, dtype: int64
No        101765
Steady         1
Name: glimepiride-pioglitazone, dtype: int64
No        101764
Steady         2
Name: metformin-rosiglitazone, dtype: int64
No        101765
Steady         1
Name: metformin-pioglitazone, dtype: int64
No    54755
Ch    47011
Name: change, dtype: int64
Yes    78363
No     23403
Name: diabetesMed, dtype: int64

"The columns to be deleted as they almost have only one value are: acetohexamide, troglitazone, glipizide-metformin, glimepiride-pioglitazone, metformin-rosiglitazone, metformin-pioglitazone, tolbutamide."

#Vérifions pour les colonnes ayant entre 3 et 5 valeurs différentes
for column in diabetic.columns:
    unique_values_count = diabetic[column].nunique()
    if 2 < unique_values_count <= 5:
        print(f"\nColumn: {column}")
        print(diabetic[column].value_counts(normalize=True) * 100)

Column: race
Caucasian          76.486788
AfricanAmerican    19.307891
Hispanic            2.047380
Other               1.513674
Asian               0.644266
Name: race, dtype: float64

Column: gender
Female             53.758623
Male               46.238429
Unknown/Invalid     0.002948
Name: gender, dtype: float64

Column: max_glu_serum
None    94.746772
Norm     2.551933
>200     1.459230
>300     1.242065
Name: max_glu_serum, dtype: float64

Column: A1Cresult
None    83.277322
>8       8.073423
Norm     4.903406
>7       3.745848
Name: A1Cresult, dtype: float64

Column: metformin
No        80.358862
Steady    18.027632
Up         1.048484
Down       0.565022
Name: metformin, dtype: float64

Column: repaglinide
No        98.487707
Steady     1.359983
Up         0.108091
Down       0.044219
Name: repaglinide, dtype: float64

Column: nateglinide
No        99.309200
Steady     0.656408
Up         0.023584
Down       0.010809
Name: nateglinide, dtype: float64

Column: chlorpropamide
No        99.915492
Steady     0.077629
Up         0.005896
Down       0.000983
Name: chlorpropamide, dtype: float64

Column: glimepiride
No        94.899082
Steady     4.588959
Up         0.321325
Down       0.190633
Name: glimepiride, dtype: float64

Column: glipizide
No        87.534147
Steady    11.158933
Up         0.756638
Down       0.550282
Name: glipizide, dtype: float64

Column: glyburide
No        89.534815
Steady     9.113063
Up         0.797909
Down       0.554213
Name: glyburide, dtype: float64

Column: pioglitazone
No        92.799167
Steady     6.854942
Up         0.229939
Down       0.115952
Name: pioglitazone, dtype: float64

Column: rosiglitazone
No        93.745455
Steady     5.994143
Up         0.174911
Down       0.085490
Name: rosiglitazone, dtype: float64

Column: acarbose
No        99.697345
Steady     0.289881
Up         0.009826
Down       0.002948
Name: acarbose, dtype: float64

Column: miglitol
No        99.962659
Steady     0.030462
Down       0.004913
Up         0.001965
Name: miglitol, dtype: float64

Column: tolazamide
No        99.961677
Steady     0.037341
Up         0.000983
Name: tolazamide, dtype: float64

Column: insulin
No        46.560737
Steady    30.313661
Down      12.005974
Up        11.119627
Name: insulin, dtype: float64

Column: glyburide-metformin
No        99.306252
Steady     0.679991
Up         0.007861
Down       0.005896
Name: glyburide-metformin, dtype: float64

Column: readmitted
NO     53.911916
>30    34.928169
<30    11.159916
Name: readmitted, dtype: float64

We decide to delete the columns that have more than 98% of the same value: repaglinide, nateglinide, chlorpropamide, acarbose, miglitol, tolazamide, glyburide-metformin. We also need to remove the rows in "sex" that are unknown".

unknown_gender_rows = diabetic[diabetic['gender'] == 'Unknown/Invalid']
unknown_gender_rows

	encounter_id	patient_nbr	race	gender	age	weight	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	...	citoglipton	insulin	glyburide-metformin	glipizide-metformin	glimepiride-pioglitazone	metformin-rosiglitazone	metformin-pioglitazone	change	diabetesMed	readmitted
30506	98634858	100695042	Other	Unknown/Invalid	[70-80)	NaN	3	1	1	1	...	No	No	No	No	No	No	No	Ch	Yes	NO
75551	226864668	60524946	NaN	Unknown/Invalid	[60-70)	NaN	1	1	7	1	...	No	No	No	No	No	No	No	No	No	NO
82573	257364294	78119847	NaN	Unknown/Invalid	[70-80)	NaN	1	22	7	8	...	No	No	No	No	No	No	No	Ch	Yes	NO

3 rows × 50 columns

#on drop les unknow race (3lignes)
diabetic = diabetic.drop(diabetic[diabetic['gender'] == 'Unknown/Invalid'].index)

diabetic_selected = diabetic.loc[:, ['weight', 'payer_code', 'medical_specialty', 'examide', 'citoglipton','acetohexamide', 'troglitazone', 'glipizide-metformin', 'glimepiride-pioglitazone','metformin-rosiglitazone', 'metformin-pioglitazone', 'tolbutamide','repaglinide', 'nateglinide', 'chlorpropamide', 'acarbose', 'miglitol', 'tolazamide','glyburide-metformin']]
print(diabetic_selected)

       weight payer_code         medical_specialty examide citoglipton  \
0         NaN        NaN  Pediatrics-Endocrinology      No          No   
1         NaN        NaN                       NaN      No          No   
2         NaN        NaN                       NaN      No          No   
3         NaN        NaN                       NaN      No          No   
4         NaN        NaN                       NaN      No          No   
...       ...        ...                       ...     ...         ...   
101761    NaN         MC                       NaN      No          No   
101762    NaN         MC                       NaN      No          No   
101763    NaN         MC                       NaN      No          No   
101764    NaN         MC           Surgery-General      No          No   
101765    NaN        NaN                       NaN      No          No   

       acetohexamide troglitazone glipizide-metformin  \
0                 No           No                  No   
1                 No           No                  No   
2                 No           No                  No   
3                 No           No                  No   
4                 No           No                  No   
...              ...          ...                 ...   
101761            No           No                  No   
101762            No           No                  No   
101763            No           No                  No   
101764            No           No                  No   
101765            No           No                  No   

       glimepiride-pioglitazone metformin-rosiglitazone  \
0                            No                      No   
1                            No                      No   
2                            No                      No   
3                            No                      No   
4                            No                      No   
...                         ...                     ...   
101761                       No                      No   
101762                       No                      No   
101763                       No                      No   
101764                       No                      No   
101765                       No                      No   

       metformin-pioglitazone tolbutamide repaglinide nateglinide  \
0                          No          No          No          No   
1                          No          No          No          No   
2                          No          No          No          No   
3                          No          No          No          No   
4                          No          No          No          No   
...                       ...         ...         ...         ...   
101761                     No          No          No          No   
101762                     No          No          No          No   
101763                     No          No          No          No   
101764                     No          No          No          No   
101765                     No          No          No          No   

       chlorpropamide acarbose miglitol tolazamide glyburide-metformin  
0                  No       No       No         No                  No  
1                  No       No       No         No                  No  
2                  No       No       No         No                  No  
3                  No       No       No         No                  No  
4                  No       No       No         No                  No  
...               ...      ...      ...        ...                 ...  
101761             No       No       No         No                  No  
101762             No       No       No         No                  No  
101763             No       No       No         No                  No  
101764             No       No       No         No                  No  
101765             No       No       No         No                  No  

[101763 rows x 19 columns]

diabetic.drop(['weight', 'payer_code', 'medical_specialty', 'examide', 'citoglipton','acetohexamide', 'troglitazone', 'glipizide-metformin', 'glimepiride-pioglitazone','metformin-rosiglitazone', 'metformin-pioglitazone', 'tolbutamide','repaglinide', 'nateglinide', 'chlorpropamide', 'acarbose', 'miglitol', 'tolazamide','glyburide-metformin'],axis=1,inplace=True)

diabetic.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 101763 entries, 0 to 101765
Data columns (total 31 columns):
 #   Column                    Non-Null Count   Dtype 
---  ------                    --------------   ----- 
 0   encounter_id              101763 non-null  int64 
 1   patient_nbr               101763 non-null  int64 
 2   race                      99492 non-null   object
 3   gender                    101763 non-null  object
 4   age                       101763 non-null  object
 5   admission_type_id         101763 non-null  int64 
 6   discharge_disposition_id  101763 non-null  int64 
 7   admission_source_id       101763 non-null  int64 
 8   time_in_hospital          101763 non-null  int64 
 9   num_lab_procedures        101763 non-null  int64 
 10  num_procedures            101763 non-null  int64 
 11  num_medications           101763 non-null  int64 
 12  number_outpatient         101763 non-null  int64 
 13  number_emergency          101763 non-null  int64 
 14  number_inpatient          101763 non-null  int64 
 15  diag_1                    101742 non-null  object
 16  diag_2                    101405 non-null  object
 17  diag_3                    100340 non-null  object
 18  number_diagnoses          101763 non-null  int64 
 19  max_glu_serum             101763 non-null  object
 20  A1Cresult                 101763 non-null  object
 21  metformin                 101763 non-null  object
 22  glimepiride               101763 non-null  object
 23  glipizide                 101763 non-null  object
 24  glyburide                 101763 non-null  object
 25  pioglitazone              101763 non-null  object
 26  rosiglitazone             101763 non-null  object
 27  insulin                   101763 non-null  object
 28  change                    101763 non-null  object
 29  diabetesMed               101763 non-null  object
 30  readmitted                101763 non-null  object
dtypes: int64(13), object(18)
memory usage: 24.8+ MB

diabetic.select_dtypes('O')

	race	gender	age	diag_1	diag_2	diag_3	max_glu_serum	A1Cresult	metformin	glimepiride	glipizide	glyburide	pioglitazone	rosiglitazone	insulin	change	diabetesMed	readmitted
0	Caucasian	Female	[0-10)	250.83	NaN	NaN	None	None	No	No	No	No	No	No	No	No	No	NO
1	Caucasian	Female	[10-20)	276	250.01	255	None	None	No	No	No	No	No	No	Up	Ch	Yes	>30
2	AfricanAmerican	Female	[20-30)	648	250	V27	None	None	No	No	Steady	No	No	No	No	No	Yes	NO
3	Caucasian	Male	[30-40)	8	250.43	403	None	None	No	No	No	No	No	No	Up	Ch	Yes	NO
4	Caucasian	Male	[40-50)	197	157	250	None	None	No	No	Steady	No	No	No	Steady	Ch	Yes	NO
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
101761	AfricanAmerican	Male	[70-80)	250.13	291	458	None	>8	Steady	No	No	No	No	No	Down	Ch	Yes	>30
101762	AfricanAmerican	Female	[80-90)	560	276	787	None	None	No	No	No	No	No	No	Steady	No	Yes	NO
101763	Caucasian	Male	[70-80)	38	590	296	None	None	Steady	No	No	No	No	No	Down	Ch	Yes	NO
101764	Caucasian	Female	[80-90)	996	285	998	None	None	No	No	Steady	No	Steady	No	Up	Ch	Yes	NO
101765	Caucasian	Male	[70-80)	530	530	787	None	None	No	No	No	No	No	No	No	No	No	NO

101763 rows × 18 columns

Calculation of the percentage of missing values (NaN).

df_null=diabetic.isnull().sum(axis=1)*100/diabetic.shape[1]
df_null

0         6.451613
1         0.000000
2         0.000000
3         0.000000
4         0.000000
            ...   
101761    0.000000
101762    0.000000
101763    0.000000
101764    0.000000
101765    0.000000
Length: 101763, dtype: float64

# "Calculation of the percentage of Nans per column
nan_percentage_by_column = diabetic.isnull().mean() * 100

print(nan_percentage_by_column)

encounter_id                0.000000
patient_nbr                 0.000000
race                        2.231656
gender                      0.000000
age                         0.000000
admission_type_id           0.000000
discharge_disposition_id    0.000000
admission_source_id         0.000000
time_in_hospital            0.000000
num_lab_procedures          0.000000
num_procedures              0.000000
num_medications             0.000000
number_outpatient           0.000000
number_emergency            0.000000
number_inpatient            0.000000
diag_1                      0.020636
diag_2                      0.351798
diag_3                      1.398347
number_diagnoses            0.000000
max_glu_serum               0.000000
A1Cresult                   0.000000
metformin                   0.000000
glimepiride                 0.000000
glipizide                   0.000000
glyburide                   0.000000
pioglitazone                0.000000
rosiglitazone               0.000000
insulin                     0.000000
change                      0.000000
diabetesMed                 0.000000
readmitted                  0.000000
dtype: float64

#replace NaN race by the most frequent

diabetic['race'].fillna(diabetic['race'].mode()[0], inplace=True)

We decide to delete the rows for which the percentage of missing values exceeds 95%. Not sure if it serves any purpose?

diabetic=diabetic.drop(df_null[df_null>95].index, axis=0)

diabetic.isnull().sum()/len(diabetic)*100

encounter_id                0.000000
patient_nbr                 0.000000
race                        0.000000
gender                      0.000000
age                         0.000000
admission_type_id           0.000000
discharge_disposition_id    0.000000
admission_source_id         0.000000
time_in_hospital            0.000000
num_lab_procedures          0.000000
num_procedures              0.000000
num_medications             0.000000
number_outpatient           0.000000
number_emergency            0.000000
number_inpatient            0.000000
diag_1                      0.020636
diag_2                      0.351798
diag_3                      1.398347
number_diagnoses            0.000000
max_glu_serum               0.000000
A1Cresult                   0.000000
metformin                   0.000000
glimepiride                 0.000000
glipizide                   0.000000
glyburide                   0.000000
pioglitazone                0.000000
rosiglitazone               0.000000
insulin                     0.000000
change                      0.000000
diabetesMed                 0.000000
readmitted                  0.000000
dtype: float64

#recording to mapping patients with discharge disposition id = 11,13,14,19,20,21

rows_to_drop = diabetic[diabetic['discharge_disposition_id'].isin([11, 13, 14, 19, 20, 21])]
print(rows_to_drop)

        encounter_id  patient_nbr             race  gender       age  \
34            293118      3327282        Caucasian  Female   [70-80)   
44            450210     80177094        Caucasian  Female   [80-90)   
64            712350    110537262  AfricanAmerican  Female   [50-60)   
78           1069716     34806888        Caucasian  Female   [50-60)   
100          1260312     38688588        Caucasian    Male   [70-80)   
...              ...          ...              ...     ...       ...   
101649     442512770     41432652        Caucasian    Male   [80-90)   
101678     442641206    141498050        Caucasian  Female   [70-80)   
101694     443139938     79902171        Caucasian    Male  [90-100)   
101704     443192642     54363114        Caucasian  Female   [80-90)   
101743     443793668     47293812        Caucasian    Male   [80-90)   

        admission_type_id  discharge_disposition_id  admission_source_id  \
34                      2                        11                    2   
44                      1                        11                    7   
64                      1                        11                    7   
78                      2                        11                    4   
100                     1                        11                    7   
...                   ...                       ...                  ...   
101649                  1                        14                    7   
101678                  3                        14                    1   
101694                  1                        13                    7   
101704                  1                        11                    7   
101743                  1                        13                    7   

        time_in_hospital  num_lab_procedures  ...  metformin  glimepiride  \
34                     5                  46  ...         No           No   
44                     7                  72  ...         No           No   
64                     4                  58  ...         No           No   
78                     2                  35  ...         No           No   
100                    9                  56  ...         No           No   
...                  ...                 ...  ...        ...          ...   
101649                11                  59  ...         No           No   
101678                 4                   2  ...         No           No   
101694                 3                  46  ...         No           No   
101704                 4                  41  ...         No           No   
101743                 3                  45  ...         No           No   

        glipizide  glyburide  pioglitazone rosiglitazone insulin change  \
34             No         No            No            No    Down     Ch   
44             No         No            No            No  Steady     No   
64             No         No            No            No    Down     Ch   
78             No         No            No            No      Up     Ch   
100            No         No            No            No  Steady     No   
...           ...        ...           ...           ...     ...    ...   
101649         No         No            No            No      Up     Ch   
101678         No         No            No            No    Down     Ch   
101694         No         No            No            No      No     No   
101704         No         No            No            No      Up     Ch   
101743         No         Up            No        Steady    Down     Ch   

        diabetesMed readmitted  
34              Yes         NO  
44              Yes         NO  
64              Yes         NO  
78              Yes         NO  
100             Yes         NO  
...             ...        ...  
101649          Yes         NO  
101678          Yes         NO  
101694           No         NO  
101704          Yes         NO  
101743          Yes         NO  

[2423 rows x 31 columns]

values_to_drop = [11, 13, 14, 19, 20, 21]

diabetic.drop(diabetic[diabetic['discharge_disposition_id'].isin(values_to_drop)].index, inplace=True)

II - Target and Visualizations

Visualization of our variables

# Drop
diabetic_dropped=diabetic.drop(["diag_1", "diag_2", "diag_3"], axis=1)

# Create a list of columns with string type
string_columns = diabetic_dropped.select_dtypes(include='object').columns.tolist()

# Iterate over the string columns
for column in string_columns:
    # Count the occurrences of each label
    counts = diabetic_dropped[column].value_counts().reset_index()
    counts.columns = ['Label', 'Count']

    # Create a pie chart using Plotly Express
    fig = px.pie(counts, values='Count', names='Label', title=column)

    # Show the pie chart
    fig.show()

Let's have a look a the readmitted column

diabetic['readmitted'].unique()

array(['NO', '>30', '<30'], dtype=object)

#Target value plot
diabetic['readmitted'].value_counts()/len(diabetic)
counts = diabetic['readmitted'].value_counts()

percentages = counts * 100 / len(diabetic)

fig, ax = plt.subplots()
colors = plt.cm.Set3.colors
percentages.sort_index(ascending=False).plot(kind='pie', ax=ax, autopct='%1.1f%%', colors=colors)

plt.show()

The goal of our study is to : "determine the early readmission of the patient within 30 days of discharge" So, only the parameters <30 and No are interesting. We can transform >30 into a no. Moreover, Since it is a binary response we'll convert our values into numeric values.

#target value 
diabetic = diabetic.replace({"NO":0,
                    "<30":1,
                    ">30":0})


print(diabetic.readmitted.value_counts())

0    88026
1    11314
Name: readmitted, dtype: int64

#target value distribution
sns.countplot(x = "readmitted", data = diabetic)
plt.title("Distribution of 'Readmitted' Values")
plt.show()

colors1 = plt.cm.Set2.colors
diabetic.readmitted.value_counts().plot.pie(autopct = "%.1f%%", colors=colors1)
plt.title("Proportion of 'Readmitted' Values")
plt.show()

#Variables historigrams
diabetic.drop(["patient_nbr"],axis=1).hist(bins=50,figsize=(20,15))
plt.show()

diabetic['gender'].unique()

array(['Female', 'Male'], dtype=object)

#gender numerisation
diabetic = diabetic.replace({"Male":0,
                    "Female":1})


print(diabetic.gender.value_counts())

1    53454
0    45886
Name: gender, dtype: int64

Correlations overview

#correlation matrix
correlation_matrix = diabetic.corr(numeric_only=True)
#heatmap
plt.figure(figsize=(10, 8))
sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', fmt='.2f', linewidths=0.5)
plt.title('Correlation Heatmap')
plt.show()

We focus on the variables that make sense and that we will use

#correlatin matrix
selected_features = ['gender','time_in_hospital','num_lab_procedures','num_procedures','number_emergency', "number_inpatient", 'num_medications', 'number_diagnoses', 'readmitted']
correlation_matrix = diabetic[selected_features].corr()
sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', fmt=".2f")
plt.title('Correlation Heatmap with Readmission Status')
plt.show()

Gender has nothing to do with beign readmitted or not or with any other variable. Correlations to explore : time in hospital and num_medication, num_lab_procedure num_procedure and num_medications number_emergency and number_inpatient

#Style
sns.set(style="whitegrid")

#Plot 1: Correlation between time_in_hospital and num_medications
plt.figure(figsize=(12, 6))
sns.scatterplot(x='time_in_hospital', y='num_medications', hue='readmitted', data=diabetic, palette='coolwarm')
plt.title('Correlation between Time in Hospital and Number of Medications')
plt.xlabel('Time in Hospital')
plt.ylabel('Number of Medications')
plt.show()

#Plot 2: Correlation between time_in_hospital and num_lab_procedures
plt.figure(figsize=(12, 6))
sns.scatterplot(x='time_in_hospital', y='num_lab_procedures', hue='readmitted', data=diabetic, palette='coolwarm')
plt.title('Correlation between Time in Hospital and Number of Lab Procedures')
plt.xlabel('Time in Hospital')
plt.ylabel('Number of Lab Procedures')
plt.show()

#Plot 3: Correlation between num_procedure and num_medications
plt.figure(figsize=(12, 6))
sns.scatterplot(x='num_procedures', y='num_medications', hue='readmitted', data=diabetic, palette='coolwarm')
plt.title('Correlation between Number of Procedures and Number of Medications')
plt.xlabel('Number of Procedures')
plt.ylabel('Number of Medications')
plt.show()

#Plot 4: Correlation between number_emergency and number_inpatient
plt.figure(figsize=(12, 6))
sns.scatterplot(x='number_emergency', y='number_inpatient', hue='readmitted', data=diabetic, palette='coolwarm')
plt.title('Correlation between Number of Emergency Visits and Number of Inpatient Visits')
plt.xlabel('Number of Emergency Visits')
plt.ylabel('Number of Inpatient Visits')
plt.show()

Plot 4 shows that : the more the patient has inpatient visits higher is the probability for the patient to be readmitted

sns.set(style="whitegrid")


columns_to_plot = ['time_in_hospital', 'num_medications', 'num_lab_procedures', 'num_procedures', 'readmitted']
plot_data = diabetic[columns_to_plot]

# Pair plot
sns.pairplot(plot_data, hue='readmitted', palette='coolwarm', markers=["o", "s"])
plt.suptitle('Pair Plot of Relevant Variables with Readmission')
plt.show()

Visualilizations of Target Variable and variables

#Count of Readmitted Cases by Race

fig = px.histogram(diabetic, x='race', color='readmitted', barmode='group', title='Count of Readmitted Cases by Race')
fig.show()

readmission status seems proportional

#count of readmitted cases by time in hospital

#cross-tabulation of readmitted cases by time in hospital
cross_tab = pd.crosstab(diabetic['time_in_hospital'], diabetic['readmitted'])

#bar chart for the count of readmitted cases by time in hospital
bar_chart = go.Bar(x=cross_tab.index, y=cross_tab[1], name='Readmitted', marker_color='red')
bar_chart_no = go.Bar(x=cross_tab.index, y=cross_tab[0], name='Not Readmitted', marker_color='blue')

data = [bar_chart, bar_chart_no]
layout = go.Layout(title='Count of Readmitted Cases by Time in Hospital',
                   xaxis=dict(title='Time in Hospital'),
                   yaxis=dict(title='Count'))

fig = go.Figure(data=data, layout=layout)
fig.show()

More time spent in hospital the less readmitted

#box plot of number of diagnoses by readmission status
fig = px.box(diabetic, x='readmitted', y='number_diagnoses', title='Box Plot of Number diagnonses by Readmission Status')
fig.show()

It seems that more diagnoses = more readmitted

#box plot of number of lab procedures by readmission status
box_plot = go.Box(x=diabetic['readmitted'], y=diabetic['num_lab_procedures'], boxpoints='all', jitter=0.3, pointpos=-1.8)

data = [box_plot]
layout = go.Layout(title='Box Plot of Number of Lab Procedures by Readmission Status',
                   xaxis=dict(title='Readmission Status'),
                   yaxis=dict(title='Number of Lab Procedures'))

fig = go.Figure(data=data, layout=layout)
fig.show()

#Pair Plot of number of diagnoses and lab procedures by Readmission Status
variables = ['number_diagnoses', 'num_lab_procedures']
index_variable = 'number_diagnoses'
target_variable = 'readmitted'

group_labels = diabetic[target_variable].unique()
scatter_matrix = ff.create_scatterplotmatrix(
    diabetic[variables], diag='histogram', index_vals=diabetic[index_variable],
    colormap='Viridis', colormap_type='cat', height=700, width=700
)

# Assign colors based on group labels
colors = {group_labels[0]: 'blue', group_labels[1]: 'red'}

for i in range(len(scatter_matrix['data'])):
    scatter_matrix['data'][i]['marker']['color'] = [colors[val] for val in diabetic[target_variable]]

scatter_matrix.update_layout(title='Pair Plot of number of diagnoses and lab procedures by Readmission Status', title_x=0.5)
scatter_matrix.show()

New Variable

diabetic['total_procedures_medications'] = diabetic['num_procedures'] + diabetic['num_medications']

fig = px.histogram(diabetic, x='total_procedures_medications', color='readmitted',
                   labels={'total_procedures_medications': 'Total Procedures/Medications'},
                   title='Distribution of Total Procedures/Medications by Readmission Status')

fig.show()

sns.countplot(x='admission_type_id', hue='readmitted', data=diabetic)
plt.title('Readmission by Admission Type ID')
plt.xlabel('Admission Type ID')
plt.ylabel('Count')
plt.show()

4,7,8 never readmitted

import seaborn as sns
import matplotlib.pyplot as plt

# Assuming 'readmitted' is a binary variable (0 or 1)
sns.set(style="whitegrid")
plt.figure(figsize=(12, 8))

# Adjusting plot width
sns.boxplot(x='age', y='time_in_hospital', hue='readmitted', data=diabetic)
plt.title('Boxplot of Time in Hospital by Age and Readmission Status')
plt.xlabel('Age')
plt.ylabel('Time in Hospital')

# Moving the legend to the right
plt.legend(loc='upper left', bbox_to_anchor=(1, 1))

plt.show()

Bokeh visualization The more the patient is at hospital the more mediaction he has

from bokeh.plotting import figure, show
from bokeh.models import ColumnDataSource, HoverTool
from bokeh.palettes import Viridis256
from bokeh.transform import linear_cmap

# Select relevant columns
selected_columns = ['time_in_hospital', 'num_medications', 'readmitted']

# Create a ColumnDataSource
source = ColumnDataSource(diabetic[selected_columns])

# Define color map
mapper = linear_cmap(field_name='readmitted', palette=Viridis256, low=min(diabetic['readmitted']), high=max(diabetic['readmitted']))

# Create the scatter plot
plot = figure(width=800, height=600, title='Scatter Plot of Time in Hospital vs. Number of Medications',
              x_axis_label='Time in Hospital', y_axis_label='Number of Medications')

scatter = plot.circle('time_in_hospital', 'num_medications', size=10, source=source, color=mapper, legend_field='readmitted', alpha=0.6)

# Add hover tool
hover = HoverTool()
hover.tooltips = [('Time in Hospital', '@time_in_hospital'), ('Number of Medications', '@num_medications'), ('Readmitted', '@readmitted')]
plot.add_tools(hover)

# Add legend
plot.legend.title = 'Readmitted'
plot.legend.label_text_font_size = '10px'

# Show the plot
show(plot)

III- Modeling and Predictions

diabetic.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 99340 entries, 0 to 101765
Data columns (total 32 columns):
 #   Column                        Non-Null Count  Dtype 
---  ------                        --------------  ----- 
 0   encounter_id                  99340 non-null  int64 
 1   patient_nbr                   99340 non-null  int64 
 2   race                          99340 non-null  object
 3   gender                        99340 non-null  int64 
 4   age                           99340 non-null  object
 5   admission_type_id             99340 non-null  int64 
 6   discharge_disposition_id      99340 non-null  int64 
 7   admission_source_id           99340 non-null  int64 
 8   time_in_hospital              99340 non-null  int64 
 9   num_lab_procedures            99340 non-null  int64 
 10  num_procedures                99340 non-null  int64 
 11  num_medications               99340 non-null  int64 
 12  number_outpatient             99340 non-null  int64 
 13  number_emergency              99340 non-null  int64 
 14  number_inpatient              99340 non-null  int64 
 15  diag_1                        99320 non-null  object
 16  diag_2                        98984 non-null  object
 17  diag_3                        97921 non-null  object
 18  number_diagnoses              99340 non-null  int64 
 19  max_glu_serum                 99340 non-null  object
 20  A1Cresult                     99340 non-null  object
 21  metformin                     99340 non-null  object
 22  glimepiride                   99340 non-null  object
 23  glipizide                     99340 non-null  object
 24  glyburide                     99340 non-null  object
 25  pioglitazone                  99340 non-null  object
 26  rosiglitazone                 99340 non-null  object
 27  insulin                       99340 non-null  object
 28  change                        99340 non-null  object
 29  diabetesMed                   99340 non-null  object
 30  readmitted                    99340 non-null  int64 
 31  total_procedures_medications  99340 non-null  int64 
dtypes: int64(16), object(16)
memory usage: 25.0+ MB

readmitted_counts = diabetic['readmitted'].value_counts()

# Create a bar chart for the distribution of values in the 'readmitted' column
bar_chart = go.Bar(x=readmitted_counts.index, y=readmitted_counts.values)
layout = go.Layout(title='Distribution of "Readmitted" Values',
                   xaxis=dict(title='Readmitted Value'),
                   yaxis=dict(title='Number of Patients'))
fig = go.Figure(data=[bar_chart], layout=layout)
fig.show()

def categorize_diagnosis(value):
    if pd.isnull(value):
        return value  # Leave null values as they are
    elif isinstance(value, str):
        try:
            value = float(value)  # Convert float strings to float
            if np.isnan(value):   # Leave NaN values as they are
                return value
            for (start, end), category in {
                (1, 139): 'Infectious and Parasitic Diseases',
                (140, 239): 'Neoplasms',
                (240, 279): 'Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders',
                (280, 289): 'Diseases of the Blood and Blood-Forming Organs',
                (290, 319): 'Mental Disorders',
                (320, 389): 'Diseases of the Nervous System and Sense Organs',
                (390, 459): 'Diseases of the Circulatory System',
                (460, 519): 'Diseases of the Respiratory System',
                (520, 579): 'Diseases of the Digestive System',
                (580, 629): 'Diseases of the Genitourinary System',
                (630, 679): 'Complications of Pregnancy, Childbirth, and the Puerperium',
                (680, 709): 'Diseases of the Skin and Subcutaneous Tissue',
                (710, 739): 'Diseases of the Musculoskeletal System and Connective Tissue',
                (740, 759): 'Congenital Anomalies',
                (760, 779): 'Certain Conditions Originating in the Perinatal Period',
                (780, 799): 'Symptoms, Signs, and Ill-Defined Conditions',
                (800, 999): 'Injury and Poisoning'
            }.items():
                if start <= value <= end:
                    return category
        except ValueError:
            return 'External Causes of Injury and Supplemental Classification'

    return value  # Return the original value if it doesn't match any category

# Example usage
import pandas as pd

# Assuming you have a DataFrame called 'df' with columns 'diag_1', 'diag_2', and 'diag_3'
diabetic['diag_1_category'] = diabetic['diag_1'].apply(categorize_diagnosis)
diabetic['diag_2_category'] = diabetic['diag_2'].apply(categorize_diagnosis)
diabetic['diag_3_category'] = diabetic['diag_3'].apply(categorize_diagnosis)
diabetic= pd.get_dummies(diabetic, columns=['diag_1_category', 'diag_2_category', 'diag_3_category'])   #  ca ca fait qu'on a pas besoin de gerer les NaN c'est cool pas de prediction a faire en plus jsp si ca vous va (ça me va)

for column in diabetic.columns:
    null_count = diabetic[column].isnull().sum()
    print(null_count)

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
21
358
1423
0
96417
84745
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

nan_rows_count = diabetic[['diag_1', 'diag_2', 'diag_3']].isnull().any(axis=1).sum()

print(f"Number of rows with NaN in diag_1, diag_2, or diag_3: {nan_rows_count}")

Number of rows with NaN in diag_1, diag_2, or diag_3: 1522

diabetic=pd.get_dummies(diabetic, columns=['race'])

diabetic['age'] = diabetic['age'].replace({
    '[0-10)': 0,
    '[10-20)': 1,
    '[20-30)': 2,
    '[30-40)': 3,
    '[40-50)': 4,
    '[50-60)': 5,
    '[60-70)': 6,
    '[70-80)': 7,
    '[80-90)': 8,
    '[90-100)': 9
})

count= diabetic['admission_type_id'].value_counts()[[5,6,8]]
count

admission_type_id
5    4785
6    5291
8     320
Name: count, dtype: int64

diabetic['admission_type_id']

0         6
1         1
2         1
3         1
4         1
         ..
101761    1
101762    1
101763    1
101764    2
101765    1
Name: admission_type_id, Length: 101763, dtype: int64

diabetic['admission_type_id'] = diabetic['admission_type_id'].replace([5, 6, 8], 0) #0 ou -10 jsp
diabetic['admission_type_id'] = diabetic['admission_type_id'].replace({4: 10, 3: 20, 2: 30, 1: 40, 7: 50}) #changer l'orde du moins grave 4 : newborn par 10, au plus grave 7 : par 14

#discharge mapping, order 

diabetic['discharge_disposition_id'] = diabetic['discharge_disposition_id'].replace({
    18: 0,
    25 : 0,
    26 : 0,
    6 : 1,
    1 : 2,
    27 : 3,
    15 : 4,
    22 : 5,
    2 : 6,
    28 : 7,
    30 : 8,
    4: 9,
    23 : 10,
    24 : 11,
    3 : 12,
    8 : 13,
    7 : 14,
    5: 15,
    9 : 16,
    10: 17,
    12 : 18,
    16 : 19,
    17 : 20,
    29 : 21
})

#admission source mapping 


diabetic['admission_source_id'] = diabetic['admission_source_id'].replace({
    9: 0,
    15: 0,
    21: 0,
    17: 0,
    20: 0,
    27: 0,
    23: 1,
    24: 1,
    11: 2,
    12: 3,
    13: 4,
    14: 5,
    8: 6,
    1: 7,
    2: 8,
    18: 9,
    3: 10,
    25: 11,
    6: 12,
    22: 13,
    4: 14,
    26: 15,
    19: 16,
    5: 17,
    7: 18,
    10: 19
})

def replace_meds(df, column_list):
    for column in column_list:
        df[column] = df[column].replace({'No': 0, 'Down': 1, 'Steady': 2, 'Up': 3})
    return df

pd.set_option('display.max_columns', None)

replace_meds(diabetic, ['metformin','glimepiride','glipizide','glyburide','pioglitazone','rosiglitazone','insulin'])

	encounter_id	patient_nbr	gender	age	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	num_lab_procedures	num_procedures	num_medications	number_outpatient	number_emergency	number_inpatient	diag_1	diag_2	diag_3	number_diagnoses	max_glu_serum	A1Cresult	metformin	glimepiride	glipizide	glyburide	pioglitazone	rosiglitazone	insulin	change	diabetesMed	readmitted	diag_1_category_Complications of Pregnancy, Childbirth, and the Puerperium	diag_1_category_Congenital Anomalies	diag_1_category_Diseases of the Blood and Blood-Forming Organs	diag_1_category_Diseases of the Circulatory System	diag_1_category_Diseases of the Digestive System	diag_1_category_Diseases of the Genitourinary System	diag_1_category_Diseases of the Musculoskeletal System and Connective Tissue	diag_1_category_Diseases of the Nervous System and Sense Organs	diag_1_category_Diseases of the Respiratory System	diag_1_category_Diseases of the Skin and Subcutaneous Tissue	diag_1_category_Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders	diag_1_category_External Causes of Injury and Supplemental Classification	diag_1_category_Infectious and Parasitic Diseases	diag_1_category_Injury and Poisoning	diag_1_category_Mental Disorders	diag_1_category_Neoplasms	diag_1_category_Symptoms, Signs, and Ill-Defined Conditions	diag_2_category_Complications of Pregnancy, Childbirth, and the Puerperium	diag_2_category_Congenital Anomalies	diag_2_category_Diseases of the Blood and Blood-Forming Organs	diag_2_category_Diseases of the Circulatory System	diag_2_category_Diseases of the Digestive System	diag_2_category_Diseases of the Genitourinary System	diag_2_category_Diseases of the Musculoskeletal System and Connective Tissue	diag_2_category_Diseases of the Nervous System and Sense Organs	diag_2_category_Diseases of the Respiratory System	diag_2_category_Diseases of the Skin and Subcutaneous Tissue	diag_2_category_Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders	diag_2_category_External Causes of Injury and Supplemental Classification	diag_2_category_Infectious and Parasitic Diseases	diag_2_category_Injury and Poisoning	diag_2_category_Mental Disorders	diag_2_category_Neoplasms	diag_2_category_Symptoms, Signs, and Ill-Defined Conditions	diag_3_category_Complications of Pregnancy, Childbirth, and the Puerperium	diag_3_category_Congenital Anomalies	diag_3_category_Diseases of the Blood and Blood-Forming Organs	diag_3_category_Diseases of the Circulatory System	diag_3_category_Diseases of the Digestive System	diag_3_category_Diseases of the Genitourinary System	diag_3_category_Diseases of the Musculoskeletal System and Connective Tissue	diag_3_category_Diseases of the Nervous System and Sense Organs	diag_3_category_Diseases of the Respiratory System	diag_3_category_Diseases of the Skin and Subcutaneous Tissue	diag_3_category_Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders	diag_3_category_External Causes of Injury and Supplemental Classification	diag_3_category_Infectious and Parasitic Diseases	diag_3_category_Injury and Poisoning	diag_3_category_Mental Disorders	diag_3_category_Neoplasms	diag_3_category_Symptoms, Signs, and Ill-Defined Conditions	race_AfricanAmerican	race_Asian	race_Caucasian	race_Hispanic	race_Other
0	2278392	8222157	Female	0	0	0	7	1	41	0	1	0	0	0	250.83	NaN	NaN	1	NaN	NaN	0	0	0	0	0	0	0	No	No	NO	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False
1	149190	55629189	Female	1	40	2	18	3	59	0	18	0	0	0	276	250.01	255	9	NaN	NaN	0	0	0	0	0	0	3	Ch	Yes	>30	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	True	False	False
2	64410	86047875	Female	2	40	2	18	2	11	5	13	2	0	1	648	250	V27	6	NaN	NaN	0	0	2	0	0	0	0	No	Yes	NO	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	True	False	False	False	False
3	500364	82442376	Male	3	40	2	18	2	44	1	16	0	0	0	8	250.43	403	7	NaN	NaN	0	0	0	0	0	0	3	Ch	Yes	NO	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False
4	16680	42519267	Male	4	40	2	18	1	51	0	8	0	0	0	197	157	250	5	NaN	NaN	0	0	2	0	0	0	2	Ch	Yes	NO	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	True	False	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
101761	443847548	100162476	Male	7	40	12	18	3	51	0	16	0	0	0	250.13	291	458	9	NaN	>8	2	0	0	0	0	0	1	Ch	Yes	>30	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False
101762	443847782	74694222	Female	8	40	9	17	5	33	3	18	0	0	1	560	276	787	9	NaN	NaN	0	0	0	0	0	0	2	No	Yes	NO	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	True	False	False	False	False
101763	443854148	41088789	Male	7	40	2	18	1	53	0	9	1	0	0	38	590	296	13	NaN	NaN	2	0	0	0	0	0	1	Ch	Yes	NO	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	True	False	False
101764	443857166	31693671	Female	8	30	12	18	10	45	2	21	0	0	1	996	285	998	9	NaN	NaN	0	0	2	0	2	0	3	Ch	Yes	NO	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	True	False	False
101765	443867222	175429310	Male	7	40	2	18	6	13	3	3	0	0	0	530	530	787	9	NaN	NaN	0	0	0	0	0	0	0	No	No	NO	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	False	True	False	False	True	False	False

101763 rows × 86 columns

diabetic['max_glu_serum']= diabetic['max_glu_serum'].replace({

    'None' : 0,
    'Norm': 10,
    '>200': 20,
    '>300': 30})
diabetic['max_glu_serum'].fillna(0, inplace=True)

diabetic['A1Cresult']= diabetic['A1Cresult'].replace({

    'None' : 0,
    'Norm': 10,
    '>7': 20,
    '>8': 30
}
)
diabetic['A1Cresult'].fillna(0, inplace=True)

diabetic['diabetesMed']=diabetic['diabetesMed'].replace({'Yes': True, 'No': False})

diabetic['change']=diabetic['change'].replace({'Ch': True, 'No': False})

for i in diabetic.columns:
    print(i+'\n')

encounter_id

patient_nbr

gender

age

admission_type_id

discharge_disposition_id

admission_source_id

time_in_hospital

num_lab_procedures

num_procedures

num_medications

number_outpatient

number_emergency

number_inpatient

diag_1

diag_2

diag_3

number_diagnoses

max_glu_serum

A1Cresult

metformin

glimepiride

glipizide

glyburide

pioglitazone

rosiglitazone

insulin

change

diabetesMed

readmitted

diag_1_category_Complications of Pregnancy, Childbirth, and the Puerperium

diag_1_category_Congenital Anomalies

diag_1_category_Diseases of the Blood and Blood-Forming Organs

diag_1_category_Diseases of the Circulatory System

diag_1_category_Diseases of the Digestive System

diag_1_category_Diseases of the Genitourinary System

diag_1_category_Diseases of the Musculoskeletal System and Connective Tissue

diag_1_category_Diseases of the Nervous System and Sense Organs

diag_1_category_Diseases of the Respiratory System

diag_1_category_Diseases of the Skin and Subcutaneous Tissue

diag_1_category_Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders

diag_1_category_External Causes of Injury and Supplemental Classification

diag_1_category_Infectious and Parasitic Diseases

diag_1_category_Injury and Poisoning

diag_1_category_Mental Disorders

diag_1_category_Neoplasms

diag_1_category_Symptoms, Signs, and Ill-Defined Conditions

diag_2_category_Complications of Pregnancy, Childbirth, and the Puerperium

diag_2_category_Congenital Anomalies

diag_2_category_Diseases of the Blood and Blood-Forming Organs

diag_2_category_Diseases of the Circulatory System

diag_2_category_Diseases of the Digestive System

diag_2_category_Diseases of the Genitourinary System

diag_2_category_Diseases of the Musculoskeletal System and Connective Tissue

diag_2_category_Diseases of the Nervous System and Sense Organs

diag_2_category_Diseases of the Respiratory System

diag_2_category_Diseases of the Skin and Subcutaneous Tissue

diag_2_category_Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders

diag_2_category_External Causes of Injury and Supplemental Classification

diag_2_category_Infectious and Parasitic Diseases

diag_2_category_Injury and Poisoning

diag_2_category_Mental Disorders

diag_2_category_Neoplasms

diag_2_category_Symptoms, Signs, and Ill-Defined Conditions

diag_3_category_Complications of Pregnancy, Childbirth, and the Puerperium

diag_3_category_Congenital Anomalies

diag_3_category_Diseases of the Blood and Blood-Forming Organs

diag_3_category_Diseases of the Circulatory System

diag_3_category_Diseases of the Digestive System

diag_3_category_Diseases of the Genitourinary System

diag_3_category_Diseases of the Musculoskeletal System and Connective Tissue

diag_3_category_Diseases of the Nervous System and Sense Organs

diag_3_category_Diseases of the Respiratory System

diag_3_category_Diseases of the Skin and Subcutaneous Tissue

diag_3_category_Endocrine, Nutritional and Metabolic Diseases, and Immunity Disorders

diag_3_category_External Causes of Injury and Supplemental Classification

diag_3_category_Infectious and Parasitic Diseases

diag_3_category_Injury and Poisoning

diag_3_category_Mental Disorders

diag_3_category_Neoplasms

diag_3_category_Symptoms, Signs, and Ill-Defined Conditions

race_AfricanAmerican

race_Asian

race_Caucasian

race_Hispanic

race_Other

diabetic = diabetic.drop(['diag_1', 'diag_2', 'diag_3'], axis=1)

Classification models

Working with the imbalanced dataset

from sklearn.model_selection import train_test_split

X = diabetic.drop('readmitted', axis=1)
Y=diabetic['readmitted']

x_train, x_test, y_train, y_test = train_test_split(X,Y, test_size = 0.3, random_state=10)

from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
scaler.fit(x_train)
x_train= scaler.transform(x_train)
x_test= scaler.transform(x_test)

Logistic Regression

from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report, recall_score, confusion_matrix, precision_score, f1_score

lg= LogisticRegression()
lg.fit(x_train, y_train)
y_pred_lg = lg.predict(x_test)

report = classification_report(y_test, y_pred_lg)
cm=confusion_matrix(y_test, y_pred_lg)
recall = recall_score(y_test, y_pred_lg)
precision = precision_score(y_test, y_pred_lg)
f1 = f1_score(y_test, y_pred_lg)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.89      1.00      0.94     26427
           1       0.43      0.01      0.03      3375

    accuracy                           0.89     29802
   macro avg       0.66      0.51      0.48     29802
weighted avg       0.84      0.89      0.84     29802
 [[26368    59]
 [ 3330    45]]
Recall: 0.013333333333333334
Precision: 0.4326923076923077
F1 Score: 0.025869502730669736

As the dataset is umbalanced, we can see that he isn't good at classifying the 1.

Logistic Regression avec GridSearch

from sklearn.model_selection import GridSearchCV
lg_params = {
    'penalty': ['l1', 'l2'],
    'C': [0.1, 1, 10],
    'solver': ['liblinear', 'saga'],
}

grid_search_lg = GridSearchCV(lg, lg_params, cv=2, n_jobs=-1, scoring='roc_auc', verbose=10)
grid_search_lg.fit(x_train, y_train)

print(grid_search_lg.best_score_)
print(grid_search_lg.best_params_)

Fitting 2 folds for each of 12 candidates, totalling 24 fits
0.6461859650012327
{'C': 0.1, 'penalty': 'l1', 'solver': 'liblinear'}

y_pred_lg= grid_search_lg.predict(x_test)
report = classification_report(y_test, y_pred_lg)
cm=confusion_matrix(y_test, y_pred_lg)
recall = recall_score(y_test, y_pred_lg)
precision = precision_score(y_test, y_pred_lg)
f1 = f1_score(y_test, y_pred_lg)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.89      1.00      0.94     26427
           1       0.45      0.01      0.03      3375

    accuracy                           0.89     29802
   macro avg       0.67      0.51      0.48     29802
weighted avg       0.84      0.89      0.84     29802
 [[26370    57]
 [ 3329    46]]
Recall: 0.01362962962962963
Precision: 0.44660194174757284
F1 Score: 0.02645198389879241

on voit que c'est mieux mais la precision à beaucoup diminué et maintenant il fait beaucoup d'erreurs sur les FP.

Toutefois,dans le contexte de l'hopital, on prefere avoir plus de FP que de FN pour des raisons de sécurité.

KNN

from sklearn.neighbors import KNeighborsClassifier
knn = KNeighborsClassifier()
knn.fit(x_train, y_train)
y_pred_knn = knn.predict(x_test)


report = classification_report(y_test, y_pred_knn)
cm=confusion_matrix(y_test, y_pred_knn)
recall = recall_score(y_test, y_pred_knn)
precision = precision_score(y_test, y_pred_knn)
f1 = f1_score(y_test, y_pred_knn)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.89      0.99      0.93     26427
           1       0.21      0.03      0.05      3375

    accuracy                           0.88     29802
   macro avg       0.55      0.51      0.49     29802
weighted avg       0.81      0.88      0.83     29802
 [[26058   369]
 [ 3278    97]]
Recall: 0.02874074074074074
Precision: 0.20815450643776823
F1 Score: 0.05050768029159074

knn = KNeighborsClassifier()
knn_params = {
    'n_neighbors': [3, 5, 7],
    'weights': ['uniform', 'distance'],
    'algorithm': ['auto', 'ball_tree'],
}

grid_search_knn = GridSearchCV(knn, knn_params, cv=2, n_jobs=-1, scoring='roc_auc', verbose=10)
grid_search_knn.fit(x_train, y_train)

print(grid_search_knn.best_score_)
print(grid_search_knn.best_params_)

Fitting 2 folds for each of 12 candidates, totalling 24 fits
0.5514742244771302
{'algorithm': 'auto', 'n_neighbors': 7, 'weights': 'distance'}

y_pred_knn= grid_search_knn.predict(x_test)
report = classification_report(y_test, y_pred_knn)
cm=confusion_matrix(y_test, y_pred_knn)
recall = recall_score(y_test, y_pred_knn)
precision = precision_score(y_test, y_pred_knn)
f1 = f1_score(y_test, y_pred_knn)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.89      0.99      0.94     26427
           1       0.24      0.02      0.03      3375

    accuracy                           0.88     29802
   macro avg       0.57      0.51      0.48     29802
weighted avg       0.81      0.88      0.84     29802
 [[26247   180]
 [ 3317    58]]
Recall: 0.017185185185185185
Precision: 0.24369747899159663
F1 Score: 0.032106282867423196

Random Forest Classifier

from sklearn.ensemble import RandomForestClassifier

rf = RandomForestClassifier()
rf.fit(x_train, y_train)
y_pred_rf = rf.predict(x_test)

report = classification_report(y_test, y_pred_lg)
cm=confusion_matrix(y_test, y_pred_lg)
recall = recall_score(y_test, y_pred_lg)
precision = precision_score(y_test, y_pred_lg)
f1 = f1_score(y_test, y_pred_lg)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.89      1.00      0.94     26427
           1       0.45      0.01      0.03      3375

    accuracy                           0.89     29802
   macro avg       0.67      0.51      0.48     29802
weighted avg       0.84      0.89      0.84     29802
 [[26370    57]
 [ 3329    46]]
Recall: 0.01362962962962963
Precision: 0.44660194174757284
F1 Score: 0.02645198389879241

rf_params = {
    'n_estimators': [150, 200, 250],
   # 'criterion': ['gini', 'entropy'],
    'max_depth': [7,10, 13],
   # 'min_samples_split': [ 10,15,20],
   #'min_samples_leaf': [ 6,8, 10],
    'max_features': ['sqrt', 'log2'],
    'max_samples': [0.5, 0.8, 1,1],
    'class_weight': ['balanced'] ,
    #'bootstrap': [True, False],
    
  
}
grid_search_p = GridSearchCV(rf, rf_params, cv=2, n_jobs=-1, scoring = 'roc_auc', verbose=10).fit(x_train, y_train)
print(grid_search_p.best_score_)
print(grid_search_p.best_params_)

Fitting 2 folds for each of 72 candidates, totalling 144 fits
0.661530892689788
{'class_weight': 'balanced', 'max_depth': 7, 'max_features': 'sqrt', 'max_samples': 0.5, 'n_estimators': 250}

y_pred_rf= grid_search_p.predict(x_test)
report = classification_report(y_test, y_pred_rf)
cm=confusion_matrix(y_test, y_pred_rf)
recall = recall_score(y_test, y_pred_rf)
precision = precision_score(y_test, y_pred_rf)
f1 = f1_score(y_test, y_pred_rf)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.92      0.66      0.77     26427
           1       0.18      0.57      0.27      3375

    accuracy                           0.65     29802
   macro avg       0.55      0.62      0.52     29802
weighted avg       0.84      0.65      0.72     29802
 [[17542  8885]
 [ 1457  1918]]
Recall: 0.5682962962962963
Precision: 0.1775432750161992
F1 Score: 0.2705600225701792

Balanced Random Forest Classifier

from imblearn.ensemble import BalancedRandomForestClassifier

rf=BalancedRandomForestClassifier()
rf.fit(x_train, y_train)
y_pred_rf = rf.predict(x_test)


report = classification_report(y_test, y_pred_rf)
cm=confusion_matrix(y_test, y_pred_rf)
recall = recall_score(y_test, y_pred_rf)
precision = precision_score(y_test, y_pred_rf)
f1 = f1_score(y_test, y_pred_rf)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:546: FutureWarning:

The default of `sampling_strategy` will change from `'auto'` to `'all'` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `'all'` to silence this warning and adopt the future behaviour.

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:558: FutureWarning:

The default of `replacement` will change from `False` to `True` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `True` to silence this warning and adopt the future behaviour.

              precision    recall  f1-score   support

           0       0.92      0.63      0.75     26427
           1       0.17      0.60      0.26      3375

    accuracy                           0.62     29802
   macro avg       0.55      0.61      0.51     29802
weighted avg       0.84      0.62      0.69     29802
 [[16576  9851]
 [ 1364  2011]]
Recall: 0.5958518518518519
Precision: 0.1695329624009442
F1 Score: 0.2639627223206668

rf = BalancedRandomForestClassifier()
rf_params = {
    'n_estimators': [150, 200, 250],
       #'criterion': ['gini', 'entropy'],
    'max_depth': [7,10, 13],
       #'min_samples_split': [ 10,15,20],
       #'min_samples_leaf': [ 6,8, 10],
    'max_features': ['sqrt', 'log2'],
    'max_samples': [0.5, 0.8, 1,1],
        #'bootstrap': [True, False],
  
}
grid_search_p = GridSearchCV(rf, rf_params, cv=2, n_jobs=-1, scoring = 'roc_auc', verbose=10).fit(x_train, y_train)
print(grid_search_p.best_score_)
print(grid_search_p.best_params_)

Fitting 2 folds for each of 72 candidates, totalling 144 fits

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:546: FutureWarning:

The default of `sampling_strategy` will change from `'auto'` to `'all'` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `'all'` to silence this warning and adopt the future behaviour.

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:558: FutureWarning:

The default of `replacement` will change from `False` to `True` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `True` to silence this warning and adopt the future behaviour.

0.6623322013585501
{'max_depth': 13, 'max_features': 'sqrt', 'max_samples': 0.8, 'n_estimators': 200}

y_pred_rf= grid_search_p.predict(x_test)
report = classification_report(y_test, y_pred_rf)
cm=confusion_matrix(y_test, y_pred_rf)
recall = recall_score(y_test, y_pred_rf)
precision = precision_score(y_test, y_pred_rf)
f1 = f1_score(y_test, y_pred_rf)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.93      0.60      0.73     26427
           1       0.17      0.63      0.26      3375

    accuracy                           0.61     29802
   macro avg       0.55      0.61      0.50     29802
weighted avg       0.84      0.61      0.68     29802
 [[15982 10445]
 [ 1265  2110]]
Recall: 0.6251851851851852
Precision: 0.1680605336519315
F1 Score: 0.26490897677338354

cm_df = pd.DataFrame(
    confusion_matrix(y_test, y_pred_rf),
    columns=grid_search_p.classes_,
    index=grid_search_p.classes_
)

sns.set(font_scale=1.4) # for label size
sns.heatmap(
    cm_df, annot=True, annot_kws={"size": 16},
    cmap='Oranges', fmt='d'
    
)

plt.title('Confusion Matrix')
plt.xlim(0, 2)
plt.ylim(0, 2)

(0.0, 2.0)

Balanced Bagging Classifier

from imblearn.ensemble import BalancedBaggingClassifier

bbc=BalancedRandomForestClassifier()
rf.fit(x_train, y_train)
y_pred_bbc = rf.predict(x_test)


report = classification_report(y_test, y_pred_bbc)
cm=confusion_matrix(y_test, y_pred_bbc)
recall = recall_score(y_test, y_pred_bbc)
precision = precision_score(y_test, y_pred_bbc)
f1 = f1_score(y_test, y_pred_bbc)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:546: FutureWarning:

The default of `sampling_strategy` will change from `'auto'` to `'all'` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `'all'` to silence this warning and adopt the future behaviour.

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:558: FutureWarning:

The default of `replacement` will change from `False` to `True` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `True` to silence this warning and adopt the future behaviour.

              precision    recall  f1-score   support

           0       0.92      0.63      0.75     26427
           1       0.17      0.59      0.26      3375

    accuracy                           0.63     29802
   macro avg       0.55      0.61      0.51     29802
weighted avg       0.84      0.63      0.70     29802
 [[16679  9748]
 [ 1372  2003]]
Recall: 0.5934814814814815
Precision: 0.17045357841885797
F1 Score: 0.26484199391775753

rf_params = {
    'n_estimators': [150, 200, 250],
    'max_depth': [7, 10, 13],
    'max_features': ['sqrt', 'log2'],
    'max_samples': [0.5, 0.8, 1],

}

grid_search_bbc = GridSearchCV(bbc, rf_params, cv=2, n_jobs=-1, scoring='roc_auc', verbose=10).fit(x_train, y_train)

print(grid_search_bbc.best_score_)
print(grid_search_bbc.best_params_)

Fitting 2 folds for each of 54 candidates, totalling 108 fits

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:546: FutureWarning:

The default of `sampling_strategy` will change from `'auto'` to `'all'` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `'all'` to silence this warning and adopt the future behaviour.

C:\Users\Izard\anaconda3\envs\PythonForDataAnalysis\lib\site-packages\imblearn\ensemble\_forest.py:558: FutureWarning:

The default of `replacement` will change from `False` to `True` in version 0.13. This change will follow the implementation proposed in the original paper. Set to `True` to silence this warning and adopt the future behaviour.

0.6621055554210042
{'max_depth': 13, 'max_features': 'sqrt', 'max_samples': 0.5, 'n_estimators': 200}

y_pred_bbc= grid_search_p.predict(x_test)
report = classification_report(y_test, y_pred_bbc)
cm=confusion_matrix(y_test, y_pred_bbc)
recall = recall_score(y_test, y_pred_bbc)
precision = precision_score(y_test, y_pred_bbc)
f1 = f1_score(y_test, y_pred_bbc)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.93      0.60      0.73     26427
           1       0.17      0.63      0.26      3375

    accuracy                           0.61     29802
   macro avg       0.55      0.61      0.50     29802
weighted avg       0.84      0.61      0.68     29802
 [[15982 10445]
 [ 1265  2110]]
Recall: 0.6251851851851852
Precision: 0.1680605336519315
F1 Score: 0.26490897677338354

Gradient Boosting Classifier

from sklearn.ensemble import GradientBoostingClassifier

gbc = GradientBoostingClassifier()
gbc.fit(x_train, y_train)
y_pred_gbc = gbc.predict(x_test)

report = classification_report(y_test, y_pred_gbc)
cm = confusion_matrix(y_test, y_pred_gbc)
recall = recall_score(y_test, y_pred_gbc)
precision = precision_score(y_test, y_pred_gbc)
f1 = f1_score(y_test, y_pred_gbc)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.89      1.00      0.94     26427
           1       0.59      0.01      0.02      3375

    accuracy                           0.89     29802
   macro avg       0.74      0.50      0.48     29802
weighted avg       0.85      0.89      0.84     29802
 [[26402    25]
 [ 3339    36]]
Recall: 0.010666666666666666
Precision: 0.5901639344262295
F1 Score: 0.0209545983701979

gbc_params = {
    'learning_rate': [0.1, 0.5, 1],
    'n_estimators': [ 200, 350,500],
    'max_depth': [3, 5, 10],
}

grid_search_gbc = GridSearchCV(gbc, gbc_params, cv=2, n_jobs=-1, scoring='roc_auc', verbose=10)
grid_search_gbc.fit(x_train, y_train)

print(grid_search_gbc.best_score_)
print(grid_search_gbc.best_params_)

Fitting 2 folds for each of 27 candidates, totalling 54 fits
0.6689080439269899
{'learning_rate': 0.1, 'max_depth': 3, 'n_estimators': 200}

y_pred_gbc= grid_search_p.predict(x_test)
report = classification_report(y_test, y_pred_gbc)
cm=confusion_matrix(y_test, y_pred_gbc)
recall = recall_score(y_test, y_pred_gbc)
precision = precision_score(y_test, y_pred_gbc)
f1 = f1_score(y_test, y_pred_gbc)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       0.92      0.66      0.77     26427
           1       0.18      0.57      0.27      3375

    accuracy                           0.65     29802
   macro avg       0.55      0.62      0.52     29802
weighted avg       0.84      0.65      0.72     29802
 [[17542  8885]
 [ 1457  1918]]
Recall: 0.5682962962962963
Precision: 0.1775432750161992
F1 Score: 0.2705600225701792

Balanced Dataset but overfited

from sklearn.utils import resample
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

X = diabetic.drop('readmitted', axis=1)
Y = diabetic['readmitted']

x_train, x_test_true, y_train, y_test_true = train_test_split(X,Y, test_size = 0.25, random_state=10)
df = pd.concat([x_train, y_train], axis=1)

# Separate majority and minority classes
df_majority = df[df['readmitted'] == 0]
df_minority = df[df['readmitted'] == 1]

# Upsample minority class
df_minority_upsampled = resample(df_minority, 
                                 replace=True,     # sample with replacement
                                 n_samples=len(df_majority),    # to match majority class
                                 random_state=1234) # reproducible results
 
# Combine majority class with upsampled minority class
df_upsampled = pd.concat([df_majority, df_minority_upsampled])
X_upsampled = df_upsampled.drop('readmitted', axis=1)
Y_upsampled = df_upsampled['readmitted']

# Display new class counts
class_counts = df_upsampled['readmitted'].value_counts() 
print(class_counts)

0    66003
1    66003
Name: readmitted, dtype: int64

x_train, x_test, y_train, y_test = train_test_split(X_upsampled, Y_upsampled, test_size = 0.25, random_state=10)

scaler = StandardScaler()

x_train= scaler.fit_transform(x_train)
x_test= scaler.transform(x_test)

from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report, recall_score, confusion_matrix, precision_score, f1_score

lg= RandomForestClassifier()
lg.fit(x_train, y_train)
y_pred_lg = lg.predict(x_test)

report = classification_report(y_test, y_pred_lg)
cm=confusion_matrix(y_test, y_pred_lg)
recall = recall_score(y_test, y_pred_lg)
precision = precision_score(y_test, y_pred_lg)
f1 = f1_score(y_test, y_pred_lg)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

              precision    recall  f1-score   support

           0       1.00      0.99      1.00     16365
           1       0.99      1.00      1.00     16637

    accuracy                           1.00     33002
   macro avg       1.00      1.00      1.00     33002
weighted avg       1.00      1.00      1.00     33002
 [[16259   106]
 [   19 16618]]
Recall: 0.9988579671815833
Precision: 0.9936618033963167
F1 Score: 0.9962531099187676

y_pred_lg2 = lg.predict(x_test_true)

report = classification_report(y_test_true, y_pred_lg2)
cm=confusion_matrix(y_test_true, y_pred_lg2)
recall = recall_score(y_test_true, y_pred_lg2)
precision = precision_score(y_test_true, y_pred_lg2)
f1 = f1_score(y_test_true, y_pred_lg2)

print(report, cm)
print("Recall:", recall)
print("Precision:", precision)
print("F1 Score:", f1)

C:\Users\lilyj\anaconda3\Lib\site-packages\sklearn\base.py:457: UserWarning:

X has feature names, but RandomForestClassifier was fitted without feature names

              precision    recall  f1-score   support

           0       0.89      1.00      0.94     22023
           1       0.00      0.00      0.00      2812

    accuracy                           0.89     24835
   macro avg       0.44      0.50      0.47     24835
weighted avg       0.79      0.89      0.83     24835
 [[22023     0]
 [ 2812     0]]
Recall: 0.0
Precision: 0.0
F1 Score: 0.0

C:\Users\lilyj\anaconda3\Lib\site-packages\sklearn\metrics\_classification.py:1469: UndefinedMetricWarning:

Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.

C:\Users\lilyj\anaconda3\Lib\site-packages\sklearn\metrics\_classification.py:1469: UndefinedMetricWarning:

Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.

C:\Users\lilyj\anaconda3\Lib\site-packages\sklearn\metrics\_classification.py:1469: UndefinedMetricWarning:

Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.

C:\Users\lilyj\anaconda3\Lib\site-packages\sklearn\metrics\_classification.py:1469: UndefinedMetricWarning:

Precision is ill-defined and being set to 0.0 due to no predicted samples. Use `zero_division` parameter to control this behavior.